]> CyberLeo.Net >> Repos - FreeBSD/FreeBSD.git/log
FreeBSD/FreeBSD.git
21 years agoLock proc while manipulating p_sigmask p_sigignore and p_siglist.
Tim J. Robbins [Fri, 14 Feb 2003 09:12:12 +0000 (09:12 +0000)]
Lock proc while manipulating p_sigmask p_sigignore and p_siglist.

21 years agoObtain proc lock around modification of p_siglist in linux_wait4().
Tim J. Robbins [Fri, 14 Feb 2003 08:59:49 +0000 (08:59 +0000)]
Obtain proc lock around modification of p_siglist in linux_wait4().

21 years agoDiff reduction to GENERIC.
Ruslan Ermilov [Fri, 14 Feb 2003 08:41:25 +0000 (08:41 +0000)]
Diff reduction to GENERIC.

21 years ago - Add a mutex for synchronizing the use of CMAP/CADDR 1 and 2.
Alan Cox [Fri, 14 Feb 2003 07:34:28 +0000 (07:34 +0000)]
 - Add a mutex for synchronizing the use of CMAP/CADDR 1 and 2.
 - Eliminate small style differences between pmap_zero_page(),
   pmap_copy_page(), etc.

21 years agoSplit the arch-specific AGP files into the appropriate files.* and do the same
Eric Anholt [Fri, 14 Feb 2003 06:33:52 +0000 (06:33 +0000)]
Split the arch-specific AGP files into the appropriate files.* and do the same
for the agp module, and add agp to the list of modules to compile for alpha.
Add an alpha_mb() to agp_flush_cache for alpha -- it's not correct but may
improve the situation, and it's what linux and NetBSD do.

21 years agoRemove an extra agp_flush_cache(). The i810 case that needs it already has it.
Eric Anholt [Fri, 14 Feb 2003 06:31:45 +0000 (06:31 +0000)]
Remove an extra agp_flush_cache().  The i810 case that needs it already has it.

21 years agoo transition from exca_write to exca_putb and read->getb. This is so that
Warner Losh [Fri, 14 Feb 2003 06:21:51 +0000 (06:21 +0000)]
o transition from exca_write to exca_putb and read->getb.  This is so that
  we can have additional different types of bridges.
o remove now bogus comment.
o Don't clear CARD_OK when we can't attach a card.
o minor style nits

# this make kldload of cardbus drivers work for me when the card is
# present on boot.

21 years agoMassive overhaul of exca to help with the isa efforts:
Warner Losh [Fri, 14 Feb 2003 06:21:18 +0000 (06:21 +0000)]
Massive overhaul of exca to help with the isa efforts:
o chip_name arrays ifdef'd out.
o use the OLDCARD-like get/put functions so we can support differnt types
  of mappings.
o Write the beggings of is this a valid exca device and introduce more
  chipset support.

# this is partially a wip, but also needed because some other cahnges I've
# made require some of these changes.

21 years agoImprovements to md5.c so that it uses getrusage to time
Mike Silbersack [Fri, 14 Feb 2003 04:48:06 +0000 (04:48 +0000)]
Improvements to md5.c so that it uses getrusage to time
benchmarks, and prints out results in float format.

21 years ago- Though I got a feedback from the originator of kern/48129 that the
Hidetoshi Shimokawa [Fri, 14 Feb 2003 03:09:59 +0000 (03:09 +0000)]
- Though I got a feedback from the originator of kern/48129 that the
previous revision fixed the panic, I found the problem exits in
another part of the function by investigating the crom dump sent by him.
The search was started in the middle of bus info block and the
routine misunderstood the EUI64 as a crom entry. This problem is fixed.

PR: kern/48129

Fix incorrect type mask included in a logical unit number and check
the validity of the lun.

21 years agoCorrect lines incorrectly added to the copyright message.
Kirk McKusick [Fri, 14 Feb 2003 00:31:06 +0000 (00:31 +0000)]
Correct lines incorrectly added to the copyright message.

Submitted by: Frank van der Linden <fvdl@wasabisystems.com>
Sponsored by:   DARPA & NAI Labs.

21 years agoAdvertize rtld(1) as ld.so(1) in manual pages world
Alexey Zelkin [Thu, 13 Feb 2003 23:07:28 +0000 (23:07 +0000)]
Advertize rtld(1) as ld.so(1) in manual pages world

21 years agoBandaid the bits that use the elink_foo() functions which are
Matthew N. Dodd [Thu, 13 Feb 2003 23:01:59 +0000 (23:01 +0000)]
Bandaid the bits that use the elink_foo() functions which are
i386 only at this point.

21 years agoFix a typo in rtld_dirname.
Alexander Kabaev [Thu, 13 Feb 2003 22:47:41 +0000 (22:47 +0000)]
Fix a typo in rtld_dirname.

21 years agoFB_INSTALL_CDEV not usable on Alpha.
David E. O'Brien [Thu, 13 Feb 2003 22:33:26 +0000 (22:33 +0000)]
FB_INSTALL_CDEV not usable on Alpha.

21 years agoFix whitespace problems with option lines.
David E. O'Brien [Thu, 13 Feb 2003 22:27:40 +0000 (22:27 +0000)]
Fix whitespace problems with option lines.

21 years agoFix the style of the SCHED_4BSD commit.
David E. O'Brien [Thu, 13 Feb 2003 22:24:44 +0000 (22:24 +0000)]
Fix the style of the SCHED_4BSD commit.

21 years agoThe protocol lock is always held in the dropafterack case, so we don't
Jeffrey Hsu [Thu, 13 Feb 2003 22:14:22 +0000 (22:14 +0000)]
The protocol lock is always held in the dropafterack case, so we don't
need to check for it at runtime.

21 years agoEnable just committed manual pages, update MLINKS
Alexey Zelkin [Thu, 13 Feb 2003 22:12:56 +0000 (22:12 +0000)]
Enable just committed manual pages, update MLINKS

21 years agoAdd AUTHORS section
Alexey Zelkin [Thu, 13 Feb 2003 22:11:19 +0000 (22:11 +0000)]
Add AUTHORS section

21 years agoFinally! Document all undocumented functions in libc_r.
Alexey Zelkin [Thu, 13 Feb 2003 22:10:10 +0000 (22:10 +0000)]
Finally! Document all undocumented functions in libc_r.

Add manual pages for following functions:

. pthread_attr_setcreatesuspend_np(3)
. pthread_main_np(3)
. pthread_multi_np(3)
. pthread_single_np(3)
. pthread_mutexattr_getkind_np(3)
. pthread_mutexattr_setkind_np(3)
. pthread_resume_all_np(3)
. pthread_resume_np(3)
. pthread_set_name_np(3)
. pthread_suspend_all_np(3)
. pthread_suspend_np(3)
. pthread_switch_add_np(3)
. pthread_switch_delete_np(3)

MFC after: 3 days

21 years agoOnly i386 has npx device.
David E. O'Brien [Thu, 13 Feb 2003 21:55:13 +0000 (21:55 +0000)]
Only i386 has npx device.

21 years agoMake the tinderbox log lines less obtrusive.
Dag-Erling Smørgrav [Thu, 13 Feb 2003 21:21:48 +0000 (21:21 +0000)]
Make the tinderbox log lines less obtrusive.

21 years agoPass KERNCONF on the command line rather than in the environment,
Dag-Erling Smørgrav [Thu, 13 Feb 2003 21:19:20 +0000 (21:19 +0000)]
Pass KERNCONF on the command line rather than in the environment,
since the command line is included in the log.

21 years agoComplete overhaul of the configuration system, allowing the user to
Dag-Erling Smørgrav [Thu, 13 Feb 2003 21:18:19 +0000 (21:18 +0000)]
Complete overhaul of the configuration system, allowing the user to
select one of multiple configurations at runtime.

21 years agoWe only provide agp drivers for the intel chipsets, move the agp
Peter Wemm [Thu, 13 Feb 2003 21:13:10 +0000 (21:13 +0000)]
We only provide agp drivers for the intel chipsets, move the agp
subdir to the i386-only section.

21 years agoDocument start_if.${ifn} scripts.
Hiroki Sato [Thu, 13 Feb 2003 20:08:11 +0000 (20:08 +0000)]
Document start_if.${ifn} scripts.

Not objected to by: -doc

21 years agoUse atomic ops to update amountpipekva. Amountpipekva represents the
Alan Cox [Thu, 13 Feb 2003 19:39:54 +0000 (19:39 +0000)]
Use atomic ops to update amountpipekva.  Amountpipekva represents the
total kernel virtual address space used by all pipes.  It is, thus, outside
the scope of any individual pipe lock.

21 years agoAdd agpreg.h missed in commit to agp_i810 (adding i85x/i86x AGP support).
Eric Anholt [Thu, 13 Feb 2003 19:24:19 +0000 (19:24 +0000)]
Add agpreg.h missed in commit to agp_i810 (adding i85x/i86x AGP support).

21 years agoBreak important implementation detail note into IMPLEMENTATION DETAILS
Alexey Zelkin [Thu, 13 Feb 2003 18:39:06 +0000 (18:39 +0000)]
Break important implementation detail note into IMPLEMENTATION DETAILS
section. Add cross reference to pthread_attr_getstack() function.

MFC after: 1 day

21 years agoAdd cross reference for pthread_attr_get_np()
Alexey Zelkin [Thu, 13 Feb 2003 18:36:48 +0000 (18:36 +0000)]
Add cross reference for pthread_attr_get_np()

21 years agoRehash MLINKS: add missing ones for recently added pthread_attr_{get,set}stack()
Alexey Zelkin [Thu, 13 Feb 2003 18:35:05 +0000 (18:35 +0000)]
Rehash MLINKS: add missing ones for recently added pthread_attr_{get,set}stack()
functions and remove links for 3 non-existent functions

21 years agoAdd appropriate MLINKS for functions documented in libgeom.3
Alexey Zelkin [Thu, 13 Feb 2003 18:27:05 +0000 (18:27 +0000)]
Add appropriate MLINKS for functions documented in libgeom.3

21 years agoFix some mdoc issues: add .Nd to NAME section, use valid .Lb argument,
Alexey Zelkin [Thu, 13 Feb 2003 18:25:48 +0000 (18:25 +0000)]
Fix some mdoc issues: add .Nd to NAME section, use valid .Lb argument,
use .Fn there appropriate

21 years agoFreeBSD 5.x does not provide libskey anymore, remove reference
Alexey Zelkin [Thu, 13 Feb 2003 18:23:13 +0000 (18:23 +0000)]
FreeBSD 5.x does not provide libskey anymore, remove reference

21 years agoAdd 'libgeom' to list of valid .Lb arguments
Alexey Zelkin [Thu, 13 Feb 2003 18:21:32 +0000 (18:21 +0000)]
Add 'libgeom' to list of valid .Lb arguments

21 years agoAcutally document how to make a LINT kernel config, besides telling all it
David E. O'Brien [Thu, 13 Feb 2003 17:55:12 +0000 (17:55 +0000)]
Acutally document how to make a LINT kernel config, besides telling all it
is gone.

21 years agoImplement dlinfo() function.
Alexander Kabaev [Thu, 13 Feb 2003 17:47:44 +0000 (17:47 +0000)]
Implement dlinfo() function.

Introdice RTLD_SELF special handle and properly process it within
dlsym() and dlinfo() functions.

The intention is to improve our compatibility with Solaris and
to make a Java port easier.

Partially submitted by: phantom

21 years agoAdd missing include files I forgot about in previous commit.
Alexander Kabaev [Thu, 13 Feb 2003 17:35:00 +0000 (17:35 +0000)]
Add missing include files I forgot about in previous commit.

21 years agoThe tcp_wrappers function `fromhost()' can fail. In such
Hajimu UMEMOTO [Thu, 13 Feb 2003 17:08:23 +0000 (17:08 +0000)]
The tcp_wrappers function `fromhost()' can fail.  In such
cases, the `struct sockaddr' will not be allocated.

Reported by: nectar
MFC after: 2 days

21 years agoRemove /usr/lib/elf from a default search path.
Alexander Kabaev [Thu, 13 Feb 2003 17:05:10 +0000 (17:05 +0000)]
Remove /usr/lib/elf from a default search path.
Move xprintf to malloc.c, it is only used there. Make static.

Submitted by: phantom

21 years agoGC the old tinderbox script.
Dag-Erling Smørgrav [Thu, 13 Feb 2003 14:27:11 +0000 (14:27 +0000)]
GC the old tinderbox script.

21 years agoRemove unnecessary EXPORT_SYMS.
Hidetoshi Shimokawa [Thu, 13 Feb 2003 13:42:19 +0000 (13:42 +0000)]
Remove unnecessary EXPORT_SYMS.

21 years agofirewire/fwohci
Hidetoshi Shimokawa [Thu, 13 Feb 2003 13:35:57 +0000 (13:35 +0000)]
firewire/fwohci
- Drain fwohci TX queue first then drain xfer queue which has not started.
- Check validity of the received packet length.
- Don't allocate too large buffer for xfer receive buf.

sbp
- Fix panic for some CROM which doesn't have a text leaf.
This could fix the PR kern/48129 but no feedback has been gotten from
the originator yet.
- Put back some M_NOWAIT flags into malloc which could be called
in interrupt context for 4-stable.

21 years agoSync to 1.114 of usbdevs
MIHIRA Sanpei Yoshiro [Thu, 13 Feb 2003 12:57:24 +0000 (12:57 +0000)]
Sync to 1.114 of usbdevs

21 years agoAdd support SUNTAC Ir-Trinity IS96U
MIHIRA Sanpei Yoshiro [Thu, 13 Feb 2003 12:55:18 +0000 (12:55 +0000)]
Add support SUNTAC Ir-Trinity IS96U

PR: 48212
Submitted by: Masaharu FUJITA <m@fjts.org>

21 years agoVarious spelling and grammar fixes
Christian Brueffer [Thu, 13 Feb 2003 12:02:51 +0000 (12:02 +0000)]
Various spelling and grammar fixes

PR: 44312
Submitted by: Chris Pepper <pepper@rockefeller.edu>
MFC after: 3 days

21 years agoPlace the logs in the tinderbox directory rather than cwd.
Dag-Erling Smørgrav [Thu, 13 Feb 2003 11:57:13 +0000 (11:57 +0000)]
Place the logs in the tinderbox directory rather than cwd.

21 years agoMove away the old log file before starting the tinderbox so we don't
Dag-Erling Smørgrav [Thu, 13 Feb 2003 10:52:46 +0000 (10:52 +0000)]
Move away the old log file before starting the tinderbox so we don't
accidentally post the old log if the tinderbox script failed to start
altogether.

21 years agoIt seems the extra precautions are no longer needed.
Dag-Erling Smørgrav [Thu, 13 Feb 2003 10:05:20 +0000 (10:05 +0000)]
It seems the extra precautions are no longer needed.

21 years ago- run the tinderbox script in verbose mode
Dag-Erling Smørgrav [Thu, 13 Feb 2003 09:59:59 +0000 (09:59 +0000)]
- run the tinderbox script in verbose mode
- use /\bStop\b/ instead of /^\*\*\* Error/ to identify errors

21 years agoAdd an XXX comment noting that getrusage() accesses p_stats->p_ru
Tim J. Robbins [Thu, 13 Feb 2003 09:53:59 +0000 (09:53 +0000)]
Add an XXX comment noting that getrusage() accesses p_stats->p_ru
and p_stats->p_cru without holding the appropriate locks.

21 years agoMissed odd address test when transcribing the Alpha version.
Peter Grehan [Thu, 13 Feb 2003 08:56:41 +0000 (08:56 +0000)]
Missed odd address test when transcribing the Alpha version.
This fixes the checksum problems seen with telnet.

21 years agoAdd Intel 85x/86x AGP support.
Eric Anholt [Thu, 13 Feb 2003 07:17:39 +0000 (07:17 +0000)]
Add Intel 85x/86x AGP support.

Submitted by: David Dawes <dawes@xfree86.org>

21 years agoMFi386
Alan Cox [Thu, 13 Feb 2003 07:03:44 +0000 (07:03 +0000)]
MFi386
 Remove kptobj.  Instead, use VM_ALLOC_NOOBJ.

21 years agoDon't bother to build ficl if NOFORTH is defined.
Benno Rice [Thu, 13 Feb 2003 04:35:04 +0000 (04:35 +0000)]
Don't bother to build ficl if NOFORTH is defined.

21 years agoAdd description about device detaching.
Hidetoshi Shimokawa [Thu, 13 Feb 2003 03:42:53 +0000 (03:42 +0000)]
Add description about device detaching.

21 years agoOops. I mis-remembered about the P4 problems. It was 5.0-DP2 that
Peter Wemm [Thu, 13 Feb 2003 02:42:06 +0000 (02:42 +0000)]
Oops.  I mis-remembered about the P4 problems.  It was 5.0-DP2 that
was shipped with DISABLE_PG_G and DISABLE_PSE, not 5.0-REL.  *blush*
Disable the code - but still leave it there in case its still lurking.

21 years agoTurn of PG_PS and PG_G for Pentium-4 cpus at boot time. This is so
Peter Wemm [Thu, 13 Feb 2003 01:52:44 +0000 (01:52 +0000)]
Turn of PG_PS and PG_G for Pentium-4 cpus at boot time.  This is so
that we can stop turning off PG_G and PG_PS globally for releases.

21 years agoAdd a 'debug.witness_trace' sysctl (and tunable) when DDB is present.
Peter Wemm [Thu, 13 Feb 2003 01:35:56 +0000 (01:35 +0000)]
Add a 'debug.witness_trace' sysctl (and tunable) when DDB is present.
This causes LOR and could-sleep messages to come with a stack trace.

21 years agoPrint "Stack backtrace:" right before dumping the backtrace. We cannot
Peter Wemm [Thu, 13 Feb 2003 01:33:59 +0000 (01:33 +0000)]
Print "Stack backtrace:" right before dumping the backtrace.  We cannot
expect end users to automatically recognize a stack trace for what it is.

21 years agoShow the paritition types that we recognize in human readable format. eg:
Peter Wemm [Thu, 13 Feb 2003 01:00:26 +0000 (01:00 +0000)]
Show the paritition types that we recognize in human readable format.  eg:
pluto2# gpt show da0
     start       end      size  contents
         0         0         1  PMBR
         1         1         1  Pri GPT header
         2        33        32  Pri GPT table
        34    401595    401562  GPT part - EFI System partition
    401596    925883    524288  GPT part - FreeBSD ufs partition
    925884   9314491   8388608  GPT part - FreeBSD swap partition
...
It'll print a plain uuid string for unrecognized types.

21 years agoAllow multiple hosts or programs to be named in program
Thomas Quinot [Thu, 13 Feb 2003 00:08:56 +0000 (00:08 +0000)]
Allow multiple hosts or programs to be named in program
or host specifications, eg:

!foo,bar
*.* /var/log/only_foo_or_bar.log

!-foo,bar
*.* /var/log/all_except_foo_or_bar.log

Reviewed by: roberto
Not objected to by: arch@

21 years agoin_pcbnotifyall() requires an exclusive protocol lock for notify functions
Jeffrey Hsu [Wed, 12 Feb 2003 23:55:07 +0000 (23:55 +0000)]
in_pcbnotifyall() requires an exclusive protocol lock for notify functions
which modify the connection list, namely, tcp_notify().

21 years agoSimplify the log accumulation code. It turns out that the trick of
Dag-Erling Smørgrav [Wed, 12 Feb 2003 21:06:19 +0000 (21:06 +0000)]
Simplify the log accumulation code.  It turns out that the trick of
looking for ^===> can give quite annoying false positives, especially
when building kernels, so drop it; the context can be inferred from
make's "Stop in /foo/bar/baz" messages anyway.

Also add a case that I'd missed the first time around (which happens
to be the common case, not the exception...)

21 years agoNO_KERNELCLEAN was a booboo.
Dag-Erling Smørgrav [Wed, 12 Feb 2003 20:29:05 +0000 (20:29 +0000)]
NO_KERNELCLEAN was a booboo.

21 years agoWelcome tbmaster, TinderboxNG's better half.
Dag-Erling Smørgrav [Wed, 12 Feb 2003 20:23:27 +0000 (20:23 +0000)]
Welcome tbmaster, TinderboxNG's better half.

21 years agoSlight cleanup; adjust verbosity; unlink the logfile before opening it
Dag-Erling Smørgrav [Wed, 12 Feb 2003 20:18:08 +0000 (20:18 +0000)]
Slight cleanup; adjust verbosity; unlink the logfile before opening it
to tickle 'tail -F'.

21 years agoo Implement C99 classification macros isfinite(), isinf(), isnan(),
Mike Barcroft [Wed, 12 Feb 2003 20:03:41 +0000 (20:03 +0000)]
o Implement C99 classification macros isfinite(), isinf(), isnan(),
  isnormal().  The current isinf() and isnan() are perserved for
  binary compatibility with 5.0, but new programs will use the macros.
o Implement C99 comparison macros isgreater(), isgreaterequal(),
  isless(), islessequal(), islessgreater(), isunordered().

Submitted by: David Schultz <dschultz@uclink.Berkeley.EDU>

21 years agoDo not do an assignment in a truth test (previous commit) or gcc gives a
Peter Wemm [Wed, 12 Feb 2003 19:34:34 +0000 (19:34 +0000)]
Do not do an assignment in a truth test (previous commit) or gcc gives a
warning which breaks builds.

cc1: warnings being treated as errors
src/sys/net/bridge.c: In function `bdg_forward':
sys/net/bridge.c:931: warning: suggest parentheses around assignment used as truth value
*** Error code 1

21 years agoAdd __unused to the declaration of yyrcsid in a more portable way.
Chad David [Wed, 12 Feb 2003 18:03:55 +0000 (18:03 +0000)]
Add __unused to the declaration of yyrcsid in a more portable way.

Discussed with: alfred, bde, jmallett, obrien

21 years agoCorrectly set bio_data in cloned children when cutting up large requests.
Tor Egge [Wed, 12 Feb 2003 16:30:46 +0000 (16:30 +0000)]
Correctly set bio_data in cloned children when cutting up large requests.

21 years agoSync to 1.113 of usbdevs
MIHIRA Sanpei Yoshiro [Wed, 12 Feb 2003 13:22:55 +0000 (13:22 +0000)]
Sync to 1.113 of usbdevs

21 years agoAdd support YAMAHA NetVolante Routers
MIHIRA Sanpei Yoshiro [Wed, 12 Feb 2003 13:20:39 +0000 (13:20 +0000)]
Add support YAMAHA NetVolante Routers
NetVolante RTA54i Broadband&ISDN Router
NetVolante RTA55i Broadband VoIP Router
NetVolante RTW65b Broadband Wireless Router
NetVolante RTW65i Broadband&ISDN Wireless Router

PR: bin/42574
Submitted by: Yoshihiko Sarumaru <mistral@imasy.or.jp>

21 years agoTeach whereintheworld to preserve TinderboxNG log messages.
Dag-Erling Smørgrav [Wed, 12 Feb 2003 09:51:47 +0000 (09:51 +0000)]
Teach whereintheworld to preserve TinderboxNG log messages.

21 years agoImplement a handle for efficient implementation of perforations in
Poul-Henning Kamp [Wed, 12 Feb 2003 09:48:27 +0000 (09:48 +0000)]
Implement a handle for efficient implementation of perforations in
lower extremities.

Setting bit 4 in debugflags (sysctl kern.geom.debugflags=16) will
allow any open to succeed on rank#1 providers.  This will generally
correspond to the physical disk devices: ad0, da0, md0 etc.

This fundamentally violates the mechanics of GEOMs autoconfiguration,
and is only provided as a debugging facility, so obviously error
reports on GEOM where this bit is or has been set will not be
accepted.

21 years agoMake "CURRENT" the default branch, and special-case it in the update
Dag-Erling Smørgrav [Wed, 12 Feb 2003 09:13:36 +0000 (09:13 +0000)]
Make "CURRENT" the default branch, and special-case it in the update
code so it results in -A rather than -rCURRENT.

21 years agoNote that graphics performance on Miata generally is better with the
Wilko Bulte [Wed, 12 Feb 2003 08:25:30 +0000 (08:25 +0000)]
Note that graphics performance on Miata generally is better with the
graphics card in a 64bit PCI slot.

21 years agoMinor correction to comment: PNIC and XIRCOM have eeprom, its just
Warner Losh [Wed, 12 Feb 2003 07:16:15 +0000 (07:16 +0000)]
Minor correction to comment: PNIC and XIRCOM have eeprom, its just
non-standard.

21 years agoImplement rman_get_device
Warner Losh [Wed, 12 Feb 2003 07:00:59 +0000 (07:00 +0000)]
Implement rman_get_device

# I though this was alredy implemented

Pointy hat on my head shown by: peter

21 years agoAdd description about kldload and security note.
Hidetoshi Shimokawa [Wed, 12 Feb 2003 06:48:45 +0000 (06:48 +0000)]
Add description about kldload and security note.

21 years agoMF-p4:
Warner Losh [Wed, 12 Feb 2003 06:11:47 +0000 (06:11 +0000)]
MF-p4:
Kill the slightly bogus #define for DECODE_PROTOTYPE
Be less verbose.  Hide most (all I hope) of the CIS
parsing behind cardbus_debug_cis (which is set with
hw.cardbus.debug_cis=1).

This doesn't fix problems with parsing, but should make cardbus
less chatty.  There appears to be some issues still with the
parsing of the CIS, but this won't fix them.

Prompted by: scottl

21 years agoMFp4:
Warner Losh [Wed, 12 Feb 2003 05:57:02 +0000 (05:57 +0000)]
MFp4:
Second part of the kldload patches for cardbus.  This makes
kldload of a driver for a device that's inserted now appears
to work.  To make it work, we only do a power cycle of the card
if there's no children drivers attached.

This likely is papering over bogosities in the power system.  The
power sequence needs to be re-written, so I'll not worry about
the papering over until the re-write.

21 years agoMFp4:
Warner Losh [Wed, 12 Feb 2003 05:54:22 +0000 (05:54 +0000)]
MFp4:
Don't reach inside of rman to r_dev.  Use rman_get_device instead.

21 years agoWhitespace nits.
Warner Losh [Wed, 12 Feb 2003 04:48:15 +0000 (04:48 +0000)]
Whitespace nits.

21 years agoRemove kptobj. Instead, use VM_ALLOC_NOOBJ.
Alan Cox [Wed, 12 Feb 2003 04:35:37 +0000 (04:35 +0000)]
Remove kptobj.  Instead, use VM_ALLOC_NOOBJ.

21 years agoBack out 1.143 and 1.144. They are no longer needed now that we start
Warner Losh [Wed, 12 Feb 2003 04:26:10 +0000 (04:26 +0000)]
Back out 1.143 and 1.144.  They are no longer needed now that we start
devd later in the boot process.  This should fix all the problems
people have had with those commits.  Diskless should be working again,
and those that mount /usr with nfs should be able to do that again too.

21 years agoStart devd after the networking stuff, but before we mount the
Warner Losh [Wed, 12 Feb 2003 04:22:40 +0000 (04:22 +0000)]
Start devd after the networking stuff, but before we mount the
critical remote systems.  This lets us run commands from devd that
aren't on the / partition.  This also means we can remove some kludges
from the networking startup that I added a while ago that caused other
problems.

There's still a race in starting devd that needs to be fixed in devd
so that things present at boot will be configured by the time devd
does daemon().  That race will be fixed later.

21 years agoNow that we don't have the NODEVFS case, we can just check to see if a
Warner Losh [Wed, 12 Feb 2003 04:12:23 +0000 (04:12 +0000)]
Now that we don't have the NODEVFS case, we can just check to see if a
device node exists directly to see if OLDCARD is compiled into the
kernel.  This eliminates the scary warning that people using NEWCARD
are seeing when they have pccard_enable=YES in their /etc/rc.conf
files.

21 years agoThis manual is called SIGNBIT(3) not FPCLASSIFY(3).
Mike Barcroft [Wed, 12 Feb 2003 03:29:39 +0000 (03:29 +0000)]
This manual is called SIGNBIT(3) not FPCLASSIFY(3).

21 years agoDocument the NOMAN option.
Tom Rhodes [Wed, 12 Feb 2003 02:30:04 +0000 (02:30 +0000)]
Document the NOMAN option.

21 years agoProperly document that syncache timer processing requires an
Jeffrey Hsu [Wed, 12 Feb 2003 00:42:12 +0000 (00:42 +0000)]
Properly document that syncache timer processing requires an
exclusive TCP protocol lock.

21 years agoImplement a bio-taskqueue to reduce number of context switches in
Poul-Henning Kamp [Tue, 11 Feb 2003 22:30:26 +0000 (22:30 +0000)]
Implement a bio-taskqueue to reduce number of context switches in
disk I/O processing.

        The intent is that the disk driver in its hardware interrupt
        routine will simply schedule the bio on the task queue with
        a routine to finish off whatever needs done.

        The g_up thread will then schedule this routine, the likely
        outcome of which is a biodone() which queues the bio on
        g_up's regular queue where it will be picked up and processed.

        Compared to the using the regular taskqueue, this saves one
        contextswitch.

Change our scheduling of the g_up and g_down queues to be water-tight,
at the cost of breaking the userland regression test-shims.

Input and ideas from:   scottl

21 years agoSay hello to Tinderbox 2.0, the choice of a new generation!
Dag-Erling Smørgrav [Tue, 11 Feb 2003 22:17:49 +0000 (22:17 +0000)]
Say hello to Tinderbox 2.0, the choice of a new generation!

21 years agoImplement C99's signbit() macro.
Mike Barcroft [Tue, 11 Feb 2003 21:56:21 +0000 (21:56 +0000)]
Implement C99's signbit() macro.

21 years agoPFIL_HOOKS optimization: check if at least one hook is present before
Sam Leffler [Tue, 11 Feb 2003 21:48:20 +0000 (21:48 +0000)]
PFIL_HOOKS optimization: check if at least one hook is present before
munging the IP header to pass to the hooks

21 years agoAdd Warner Losh quote.
Wes Peters [Tue, 11 Feb 2003 21:38:02 +0000 (21:38 +0000)]
Add Warner Losh quote.
Submitted by: Alexandr Kovalenko <never@nevermind.kiev.ua>

21 years agoAnnounce our ability to do DFLTPHYS sized transfers.
Poul-Henning Kamp [Tue, 11 Feb 2003 21:30:28 +0000 (21:30 +0000)]
Announce our ability to do DFLTPHYS sized transfers.

21 years agoAnnounce our ability to do MAXPHYS transfers.
Poul-Henning Kamp [Tue, 11 Feb 2003 21:29:53 +0000 (21:29 +0000)]
Announce our ability to do MAXPHYS transfers.

21 years agoAdvertise MAXPHYS upwards, we will split as necessary before we get to the
Poul-Henning Kamp [Tue, 11 Feb 2003 21:24:25 +0000 (21:24 +0000)]
Advertise MAXPHYS upwards, we will split as necessary before we get to the
bottom of things.